Position 1
Position 2
Position 3
More
Hmm. I think that this will be hard to figure out. If it is evaluating the prospects of a kingside attack starting with g4, perhaps there are lines it sees where the king becomes exposed to an attack on c1.
Who knows?
Good question. My best guess is, that it value king safety (normally better at b1 than c1) and the protection of the a2-pawn. But I also think it (at the level of 20 ply) really doesn't know what else to do. Don't trust quick engine evaluations in these kind of non tactical positions.
@1
'Amateurs are often puzzled by this apparent loss of time.
Actually it is a handy defensive move... One never knows when lightning will strike!' - Fischer
Kb1 protects pawn a2, removes the king from the c-file and from the diagonal c1-h6.
I am trying to understand why the engine is enthousiastic about Kb1 i
In one of the positions, Kb1 has an evaluation score 0.2 higher than tthe second best move. In the other 2 positions, Kb1 is not even the engine's top choice.
"Enthusiastic", sure.
I'm too lazy to look at the analysis stuff lol but I know that kb1 is common for white (or kb8 for black) when they queenside castle so that the king protects the 3 pawns
Also, (not in these positions) Its a move that entails counterattacks if Qa5 with the suprising Nd5! allowing an intermediate check after the exchange of queens, in f6 or d7.
Kb1 is obligatory in such long castle positions with exposed dark-squared diagonal at a high level!
The reason is mostly to prevent checks. This loses you an entire tempo which at a high level can be devastating, especially in these closed positions where it's all about who gets the attacking initiative.
It's the same reason why in many short castling positions engines love h3 or h4+g3. They see down the line.
And yes also protect the pawn, because that gains you an active knight.
But as you can see, it's not a priority move as of yet because there are other points to unlock first.
But as soon as it makes sense (and you'd like the opponent to err first 
) play Kb1. The proof is in the pudding, said the man with a flan.
I am trying to understand why the engine is enthousiastic about Kb1 in the three resembling positions below. Can anyone help?